CN114519529A - Enterprise credit rating method, device and medium based on convolution self-encoder - Google Patents
Enterprise credit rating method, device and medium based on convolution self-encoder Download PDFInfo
- Publication number
- CN114519529A CN114519529A CN202210158895.6A CN202210158895A CN114519529A CN 114519529 A CN114519529 A CN 114519529A CN 202210158895 A CN202210158895 A CN 202210158895A CN 114519529 A CN114519529 A CN 114519529A
- Authority
- CN
- China
- Prior art keywords
- fields
- enterprise
- data
- encoder
- entering
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000003064 k means clustering Methods 0.000 claims abstract description 19
- 238000012545 processing Methods 0.000 claims abstract description 19
- 238000004140 cleaning Methods 0.000 claims abstract description 14
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 9
- 238000011156 evaluation Methods 0.000 claims description 13
- 238000003860 storage Methods 0.000 claims description 11
- 238000012549 training Methods 0.000 claims description 11
- 238000012216 screening Methods 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 7
- 230000001502 supplementing effect Effects 0.000 claims description 2
- 238000011161 development Methods 0.000 abstract description 5
- 230000019771 cognition Effects 0.000 abstract description 3
- 238000004458 analytical method Methods 0.000 abstract description 2
- 238000013135 deep learning Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 12
- 238000004590 computer program Methods 0.000 description 7
- 230000006870 function Effects 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000012797 qualification Methods 0.000 description 2
- 208000027418 Wounds and injury Diseases 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000009193 crawling Effects 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 208000014674 injury Diseases 0.000 description 1
- 239000000047 product Substances 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000007115 recruitment Effects 0.000 description 1
- 238000012502 risk assessment Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/03—Credit; Loans; Processing thereof
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- General Business, Economics & Management (AREA)
- Development Economics (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Probability & Statistics with Applications (AREA)
- Educational Administration (AREA)
- Technology Law (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application discloses a method, equipment and a medium for enterprise credit rating based on a convolution self-encoder, wherein the method comprises the following steps: acquiring an enterprise portrait dimension table, and acquiring data according to preset fields contained in the enterprise portrait dimension table to obtain corresponding multiple pieces of enterprise data; performing data selection and data cleaning processing on enterprise data to obtain a model entering field and corresponding model entering enterprise data; according to the model entering fields, inputting the corresponding model entering enterprise data to a network model based on a convolution self-encoder respectively to obtain reconstruction data corresponding to the model entering fields respectively; and according to the module-entering field, performing K-means clustering on the corresponding reconstruction data for preset times respectively to obtain corresponding clustering results, and determining the credit rating of the enterprise according to the clustering results. Deep learning analysis is carried out by combining a convolution self-encoder and a K-means algorithm, so that the credit rating of an enterprise is obtained, preliminary cognition on the enterprise is favorably established, and comprehensive control on the enterprise operation and development condition risk points is established.
Description
Technical Field
The application relates to the technical field of machine learning, in particular to a method, equipment and medium for enterprise credit rating based on a convolution self-encoder.
Background
With the development of technologies such as big data, machine learning and artificial intelligence, the traditional financial operation service mode is changed greatly. With the rapid development of internet finance, it becomes increasingly important to perform risk assessment or credit rating on financial big data of various enterprises.
The comprehensive interpretation of the financial big data to the enterprise is favorable for banks or other financial institutions to establish preliminary cognition on the enterprise, establish comprehensive control over the enterprise operation and development condition risk points and facilitate the banks or other financial institutions to perform corresponding loan or interest adjustment operations on the enterprise. How to combine machine learning, artificial intelligence technology and financial big data to construct a corresponding enterprise credit rating model is a problem to be solved urgently.
Disclosure of Invention
In order to solve the above problems, the present application provides a method, an apparatus, and a medium for enterprise credit rating based on a convolutional auto-encoder, including:
in a first aspect, the present application provides an enterprise credit rating method based on a convolutional auto-encoder, including: acquiring an enterprise portrait dimension table, and acquiring data according to preset fields contained in the enterprise portrait dimension table to obtain corresponding multiple pieces of enterprise data; performing data selection and data cleaning processing on the enterprise data to obtain a model entering field and corresponding model entering enterprise data; according to the model entering fields, inputting the corresponding model entering enterprise data to a network model based on a convolution self-encoder respectively to obtain reconstruction data corresponding to the model entering fields respectively; and according to the module-entering field, performing K-means clustering on the corresponding reconstruction data for preset times respectively to obtain corresponding clustering results, and determining the credit rating of the enterprise according to the clustering results.
In one example, performing data selection and data cleaning processing on the enterprise data to obtain an entry module field and corresponding entry module enterprise data specifically includes: deleting the enterprise data with the missing value larger than a first preset threshold value and the corresponding preset fields according to the preset fields and the corresponding enterprise data to obtain the deleted fields and the deleted enterprise data corresponding to the deleted fields; and selecting a preset number of fields and corresponding enterprise data from the deleted fields according to a pre-stored selection rule to obtain the cleaned fields and the cleaned enterprise data corresponding to the cleaned fields.
In one example, after selecting a preset number of fields and corresponding enterprise data according to a pre-stored selection rule in the deleted fields to obtain cleaned fields and cleaned enterprise data corresponding to the cleaned fields, the method further includes: selecting a plurality of fields with the correlation larger than a second preset threshold value from the cleaned fields, selecting the fields with the most corresponding enterprise data from the fields with the correlation larger than the second preset threshold value for reservation, and deleting the rest fields and the corresponding enterprise data; determining a field corresponding to the enterprise data with the missing value larger than a third preset threshold value in the rest fields, carrying out mean processing on the enterprise data with the missing value larger than the third preset threshold value, and supplementing the enterprise data with the missing value larger than the third preset threshold value to the corresponding field; and obtaining the model entering field and the corresponding model entering enterprise data.
In one example, in the deleted fields, a preset number of fields and corresponding enterprise data are selected according to a pre-stored selection rule, so as to obtain cleaned fields and cleaned enterprise data corresponding to the cleaned fields, which specifically includes: selecting a plurality of fields with the correlation larger than a fourth preset threshold value from the deleted fields, selecting the fields with the maximum corresponding enterprise data from the fields with the correlation larger than the fourth preset threshold value for reservation, and deleting the rest fields and the corresponding enterprise data; screening the deleted fields to obtain a first preset number of times fields, wherein the times fields are fields reflecting the statistics times; screening the deleted fields to obtain a second preset number of time fields, wherein the time fields are fields reflecting the farthest time or the latest time; screening the deleted fields to obtain a third preset number of numerical fields, wherein the numerical fields reflect fields needing numerical operation; screening the deleted fields to obtain a fourth preset number of simple fields, wherein the simple fields reflect fields which do not need numerical operation; and obtaining the cleaned field and the cleaned enterprise data corresponding to the cleaned field.
In one example, before the corresponding modelled enterprise data are respectively input to a network model based on a convolutional auto-encoder according to the modelled field, and reconstructed data respectively corresponding to the modelled field are obtained, the method further includes: according to the model entering field, inputting the corresponding model entering enterprise data into a network model to be trained based on a convolution self-encoder for training; according to the training result, adjusting parameters of the network model to be trained based on the convolution self-encoder; repeating training until the mean square error between the output result of the network model of the convolutional self-encoder to be trained and the input model-entering enterprise data is smaller than a fifth preset threshold; and obtaining a network model based on the convolution self-encoder.
In one example, according to a training result, performing parameter adjustment on the network model to be trained based on the convolutional auto-encoder specifically includes: and adjusting the learning rate of the network model to be trained based on the convolution self-encoder, the network structure and the numerical value corresponding to the feature map.
In one example, according to the module-entering field, performing K-means clustering on the corresponding reconstruction data for preset times respectively to obtain corresponding clustering results, and determining an enterprise credit rating according to the clustering results, specifically including: according to the module entering field, performing K-means clustering on the corresponding reconstruction data for preset times respectively to obtain corresponding clustering results, wherein the clustering results comprise: a cluster label, a center value corresponding to the cluster label; and determining a cluster label with the maximum central value and a module-entering field corresponding to the cluster label, and taking the cluster label with the maximum central value as an evaluation parameter of the corresponding module-entering field.
In one example, after determining a cluster index with a maximum center value and a modulus-entering field corresponding to the cluster index, and taking the cluster index with the maximum center value as an evaluation parameter of the corresponding modulus-entering field, the method further includes: and obtaining respectively corresponding evaluation parameters according to the mould-entering fields, and calculating the respectively corresponding evaluation parameters through a rating algorithm to obtain the enterprise credit rating.
On the other hand, the application also provides enterprise credit rating equipment based on the convolution self-encoder, which comprises the following components: at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to: acquiring an enterprise portrait dimension table, and acquiring data according to preset fields contained in the enterprise portrait dimension table to obtain corresponding multiple pieces of enterprise data; performing data selection and data cleaning processing on the enterprise data to obtain a model entering field and corresponding model entering enterprise data; according to the model entering fields, inputting the corresponding model entering enterprise data to a network model based on a convolution self-encoder respectively to obtain reconstruction data corresponding to the model entering fields respectively; and according to the module-entering field, performing K-means clustering on the corresponding reconstruction data for preset times respectively to obtain corresponding clustering results, and determining the credit rating of the enterprise according to the clustering results.
In another aspect, the present application further provides a non-volatile computer storage medium storing computer-executable instructions configured to: acquiring an enterprise portrait dimension table, and acquiring data according to preset fields contained in the enterprise portrait dimension table to obtain corresponding multiple pieces of enterprise data; performing data selection and data cleaning processing on the enterprise data to obtain a model entering field and corresponding model entering enterprise data; according to the model entering fields, inputting the corresponding model entering enterprise data to a network model based on a convolution self-encoder respectively to obtain reconstruction data corresponding to the model entering fields respectively; and according to the module-entering field, performing K-means clustering on the corresponding reconstruction data for preset times respectively to obtain corresponding clustering results, and determining the credit rating of the enterprise according to the clustering results.
The enterprise credit rating method, the device and the medium based on the convolution self-encoder, which are provided by the application, can bring the following beneficial effects: by combining the convolution self-encoder and the K-means algorithm, deep learning and deep analysis are carried out on various fields of the enterprise, and then corresponding enterprise credit rating is obtained, so that preliminary cognition on the enterprise is favorably established by banks or other financial institutions, comprehensive control on enterprise operation and development condition risk points is established, and the corresponding loan or adjustment operation is conveniently carried out on the enterprise by the banks or other financial institutions.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the application and together with the description serve to explain the application and not to limit the application. In the drawings:
FIG. 1 is a schematic flowchart of an enterprise credit rating method based on a convolution self-encoder according to an embodiment of the present application;
FIG. 2(a) is a schematic diagram of the first K-means clustering of reconstructed data in the embodiment of the present application;
FIG. 2(b) is a schematic diagram of the second K-means clustering of the reconstructed data in the embodiment of the present application;
FIG. 2(c) is a schematic diagram of a third K-means cluster of reconstructed data in the embodiment of the present application;
FIG. 2(d) is a diagram illustrating a fourth K-means clustering of reconstructed data according to an embodiment of the present application;
FIG. 2(e) is a diagram illustrating a fifth K-means clustering of reconstructed data according to an embodiment of the present application;
fig. 3 is a schematic diagram of an enterprise credit rating device based on a convolution self-encoder in the embodiment of the application.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the technical solutions of the present application will be described in detail and completely with reference to the following specific embodiments of the present application and the accompanying drawings. It should be apparent that the described embodiments are only some of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
First, it should be noted that the enterprise credit rating method based on the convolutional auto-encoder described in the present application may be stored in a corresponding terminal in the form of a program or an algorithm, and in order to support the normal operation of the system, the terminal should have corresponding hardware, such as a processor, a memory, a communication module, and the like, so as to further support the system, that is, the program and the algorithm. In addition, the terminal can also interact with a remote server, and the hardware and corresponding computing power of the remote server are utilized to realize the same functions. In addition, the form of the terminal includes, but is not limited to: personal computers, smart phones, tablet computers or other terminal equipment with corresponding functions. Users can log in the system in the forms of the system, APP or WEB page and the like so as to control, allocate and monitor corresponding functions in the system.
The technical solutions provided by the embodiments of the present application are described in detail below with reference to the accompanying drawings.
As shown in fig. 1, an enterprise credit rating method based on a convolutional self-encoder provided in an embodiment of the present application includes:
s101: and acquiring an enterprise portrait dimension table, and acquiring data according to preset fields contained in the enterprise portrait dimension table to obtain corresponding multiple pieces of enterprise data.
Specifically, different enterprise portrait dimension tables can be set for different enterprises, the enterprise portrait dimension tables can be pre-stored in a memory or a database where the system interacts with, and the system can call the enterprise portrait dimension tables in the memory or the database.
In this embodiment, the enterprise portrait dimension table may include: enterprise background, enterprise stability, operational capability, enterprise credibility, judicial risk, operational risk, enterprise credit increase, credit risk, technological innovation capability and other dimensions.
Meanwhile, the enterprise context includes, but is not limited to, the following preset fields: registered capital, number of workers, dates of establishment, business type, business category, business hours, zip code.
Enterprise stability includes, but is not limited to, the following preset fields: enterprise change times and stock right change times.
The business capabilities include, but are not limited to, the following preset fields: the number of online stores, the number of branches of an enterprise, the number of enterprise investments, the number of external guarantees, the number of external investments, the number of enterprise bid wining and the number of recruitment records.
The reputation of the enterprise includes, but is not limited to, the following preset fields: the method comprises the following steps of enterprise insurance participation duration, unit business insurance accumulated owing amount, unit participation worker basic medical insurance accumulated owing amount, unit participation industrial injury insurance accumulated owing amount, unit participation town worker basic old insurance accumulated owing amount, unit participation production insurance accumulated owing amount and qualification rate of enterprise products subjected to spot check.
Judicial risks include, but are not limited to, the following preset fields: the number of the appellations, the number of the appellations and the target of the litigation amount.
Operational risks include, but are not limited to, the following preset fields: whether the enterprise is listed with abnormal operation or not, whether the enterprise has administrative penalty records or not, the number of the enterprise equity qualification records, the accumulated amount of owed tax of the enterprise and whether the company has personal administrative penalty or not.
Enterprise trust includes, but is not limited to, the following preset fields: whether the trade mark is listed as a business name of coming, whether the trade mark is listed as a famous trade mark, whether the trade mark is listed as a contract for credit, and whether the trade mark belongs to a local encouragement policy list.
Credit risk includes, but is not limited to, the following preset fields: whether the enterprise is listed in a credit loss blacklist or not and whether the enterprise is a credit loss enterprise of the department of industry and commerce.
The technological innovation capabilities include, but are not limited to, the following preset fields: the times of registration of copyright of enterprise software, the times of application of enterprise patent, and whether the enterprise is used for intellectual property of domain name.
Furthermore, the system can acquire data according to the preset fields, and the acquisition mode includes but is not limited to: inquiry of corresponding official website, search and crawling of internet keywords, telephone inquiry and the like.
Further, the system can obtain a plurality of pieces of enterprise data corresponding to the preset fields.
S102: and performing data selection and data cleaning processing on the enterprise data to obtain the model entering field and the corresponding model entering enterprise data.
Specifically, the system deletes the enterprise data with the missing value larger than the first preset threshold value and the corresponding preset field according to the preset field and the corresponding plurality of enterprise data to obtain the deleted field and the deleted enterprise data corresponding to the deleted field.
And then, the system selects a preset number of fields and corresponding enterprise data from the deleted fields according to a pre-stored selection rule to obtain the cleaned fields and the cleaned enterprise data corresponding to the cleaned fields.
The field selection through the selection rule specifically includes: and selecting a plurality of fields with the correlation larger than a fourth preset threshold value from the deleted fields, selecting the fields with the most corresponding enterprise data from the fields with the correlation larger than the fourth preset threshold value for reservation, and deleting the rest fields and the corresponding enterprise data.
It should be noted that the correlation here can be identified by a semantic identification algorithm in the system, and when there is correlation larger than a fourth preset threshold in a plurality of fields, it is indicated that the semantics are very similar and the data that can be reflected are also similar, so that it is not necessary to retain all of the fields, and it is sufficient to retain the field with the most enterprise data.
Furthermore, the system screens the deleted fields to obtain a first preset number of times field, wherein the times field is a field reflecting the statistical times. For example, the number of complaints field, i.e., the number of times field, corresponds to 10 enterprise data.
Furthermore, the system screens the deleted fields to obtain a second preset number of time fields, wherein the time fields are fields reflecting the farthest or the latest time. For example, the business time field is 3 corresponding to the enterprise data, and the business time field is a time field.
And then, the system screens the deleted fields to obtain a third preset number of numerical fields, wherein the numerical fields are fields reflecting numerical operation. For example, the zip code field corresponds to 250002 business data, and the numerical value corresponding to the data is too large to be directly used as the model-entering business data, so that corresponding calculation is required, for example, 250002 is divided by 100000, and the decimal is used as the model-entering business data. Here, a field requiring simple calculation processing is a numerical value field. It should be further noted that the selected numeric field needs to be operated according to the operation rule corresponding to the field, and then is used as the final enterprise data.
Furthermore, the system screens the deleted fields to obtain a fourth preset number of simple fields, wherein the simple fields are fields which reflect that numerical operation is not required. That is, unlike the numerical value field described above, a field in which an operation is not required is used.
It should be noted that in the process of screening or selecting the deleted fields, the selected fields are deleted again from the originally deleted fields, so as to avoid the situation that the same field is selected multiple times.
By the technical scheme, the cleaned field and the cleaned enterprise data corresponding to the cleaned field can be obtained. Through data cleaning, the selected fields and enterprise data can be guaranteed to have better representativeness and characteristics.
In addition, after the system selects a preset number of fields and corresponding enterprise data according to a pre-stored selection rule in the deleted fields to obtain the cleaned fields and the cleaned enterprise data corresponding to the cleaned fields, the system may further:
the system selects a plurality of fields with the correlation larger than a second preset threshold value from the cleaned fields, selects the field with the most corresponding enterprise data from the fields with the correlation larger than the second preset threshold value for reservation, and deletes the rest fields and the corresponding enterprise data.
It should be noted that the correlation selection here is similar to the above technical solution, and a specific implementation manner is not described herein again, and meanwhile, since the correlation selection here is the second correlation selection, the second preset threshold is smaller than the fourth preset threshold, that is, the correlation selected for the second time is smaller than the correlation in the foregoing, which is beneficial to more data simplification.
And then, the system determines a field corresponding to the enterprise data with the missing value larger than the third preset threshold value in the remaining fields, performs mean processing on the enterprise data with the missing value larger than the third preset threshold value, and supplements the enterprise data to the corresponding field.
In addition, it should be noted that, in the data collection process of the system, some collected enterprise data may not have corresponding preset fields, and at this time, the system may construct new fields, such as zip codes, according to the enterprise data and the description of the web pages of the enterprise data.
Furthermore, in the embodiment of the present application, through the data selection and the data cleaning, twelve modulo fields can be finally obtained, as shown in the following table:
it should be noted that, the order of the field description and the modulo field in the above table is in one-to-one correspondence, and in the final clustering result obtained in the following, the result display is performed by the same field description as that here.
S103: and respectively inputting the corresponding model-entering enterprise data to a network model based on a convolution self-encoder according to the model-entering fields to obtain the reconstruction data respectively corresponding to the model-entering fields.
Specifically, before this, the system needs to train to obtain a network model based on a convolution self-encoder.
The method specifically comprises the following steps:
and respectively inputting the corresponding model-entering enterprise data into a network model to be trained based on a convolution self-encoder according to the model-entering field by the system for training.
And then, the system adjusts parameters of the network model to be trained based on the convolution self-encoder according to the training result.
The parameter adjustment specifically comprises: and adjusting the learning rate of the network model to be trained based on the convolution self-encoder, the network structure and the numerical value corresponding to the feature map.
And further, repeating the training until the mean square error between the output result of the network model of the convolutional self-encoder to be trained and the input model-entering enterprise data is smaller than a fifth preset threshold value.
Further, the system obtains a network model based on the convolutional autoencoder.
S104: and according to the module-entering field, performing K-means clustering on the corresponding reconstruction data for preset times respectively to obtain corresponding clustering results, and determining the credit rating of the enterprise according to the clustering results.
Specifically, as shown in fig. 2(a) -2 (e), in the embodiment of the present application, five consecutive K-means clusters are performed on the reconstructed data corresponding to the enterprise registered capital field, so as to obtain a corresponding clustering result, that is, fig. 2(e) is used as a final clustering result, and corresponding numerical data is collected.
Furthermore, five times of K-means clustering are respectively carried out on the reconstruction data respectively corresponding to all the in-mode fields, and clustering results shown in the following table are obtained.
Further, the system obtains a clustering result, the clustering result comprising: cluster index, center value corresponding to cluster index.
In the above table, 0, 1, and 2 are three cluster numbers, and the numerical value below the cluster number is the center value.
And then, the system determines the cluster label with the maximum central value and the module-entering field corresponding to the cluster label, and takes the cluster label with the maximum central value as the evaluation parameter of the corresponding module-entering field.
For example, the module entry field corresponding to recrgap in the above table is the registered capital, and in the cluster corresponding to recrgap, the value with the largest central value is 0.40378964, and the cluster number corresponding to 0.40378964 is 2, at which time 2 is used as the module entry field, i.e. the evaluation parameter of the registered capital.
And then, the system obtains the corresponding evaluation parameters according to the in-mold field, and calculates the corresponding evaluation parameters through a rating algorithm to obtain the enterprise credit rating.
In the embodiment of the present application, when the maximum central value appears in cluster 1, it indicates that the credit rating of the enterprise corresponding to the modular field is high, when the maximum central value appears in cluster 0, it indicates that the credit rating of the enterprise corresponding to the modular field is medium, and when the maximum central value appears in cluster 2, it indicates that the credit rating of the enterprise corresponding to the modular field is poor.
In addition, the rating algorithm can perform corresponding calculation by accumulating or adding coefficients to a plurality of evaluation parameters to obtain the final comprehensive credit rating of the enterprise.
In one embodiment, as shown in fig. 3, the present application further provides an enterprise credit rating apparatus based on a convolutional auto-encoder, comprising:
at least one processor; and the number of the first and second groups,
a memory communicatively coupled to the at least one processor; wherein,
the memory stores instructions executable by the at least one processor to cause the at least one processor to perform instructions for:
Acquiring an enterprise portrait dimension table, and acquiring data according to preset fields contained in the enterprise portrait dimension table to obtain a plurality of corresponding enterprise data;
performing data selection and data cleaning processing on the enterprise data to obtain a model entering field and corresponding model entering enterprise data;
according to the model entry fields, inputting the corresponding model entry enterprise data to a network model based on a convolution self-encoder respectively to obtain reconstructed data corresponding to the model entry fields respectively;
and according to the module-entering field, respectively carrying out K-means clustering on the corresponding reconstruction data for preset times to obtain corresponding clustering results, and determining the credit rating of the enterprise according to the clustering results.
In one embodiment, the present application further provides a non-transitory computer storage medium storing computer-executable instructions configured to:
acquiring an enterprise portrait dimension table, and acquiring data according to preset fields contained in the enterprise portrait dimension table to obtain corresponding multiple pieces of enterprise data;
performing data selection and data cleaning processing on the enterprise data to obtain a model entering field and corresponding model entering enterprise data;
According to the model entry fields, inputting the corresponding model entry enterprise data to a network model based on a convolution self-encoder respectively to obtain reconstructed data corresponding to the model entry fields respectively;
and according to the module-entering field, respectively carrying out K-means clustering on the corresponding reconstruction data for preset times to obtain corresponding clustering results, and determining the credit rating of the enterprise according to the clustering results.
The embodiments in the present application are described in a progressive manner, and the same and similar parts among the embodiments can be referred to each other, and each embodiment focuses on differences from other embodiments. In particular, for the device and media embodiments, the description is relatively simple, as it is substantially similar to the method embodiments, and reference may be made to some description of the method embodiments for relevant points.
The device and the medium provided by the embodiment of the application correspond to the method one by one, so the device and the medium also have the beneficial technical effects similar to the corresponding method.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and so forth) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
The memory may include forms of volatile memory in a computer readable medium, Random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of a computer-readable medium.
Computer-readable media, including both permanent and non-permanent, removable and non-removable media, may implement the information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), Digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in the process, method, article, or apparatus that comprises the element.
The above description is only an example of the present application and is not intended to limit the present application. Various modifications and changes may occur to those skilled in the art to which the present application pertains. Any modification, equivalent replacement, improvement or the like made within the spirit and principle of the present application shall be included in the scope of the claims of the present application.
Claims (10)
1. The enterprise credit rating method based on the convolution self-encoder is characterized by comprising the following steps:
acquiring an enterprise portrait dimension table, and acquiring data according to preset fields contained in the enterprise portrait dimension table to obtain corresponding multiple pieces of enterprise data;
performing data selection and data cleaning processing on the enterprise data to obtain a model entering field and corresponding model entering enterprise data;
according to the model entry fields, inputting the corresponding model entry enterprise data to a network model based on a convolution self-encoder respectively to obtain reconstructed data corresponding to the model entry fields respectively;
and according to the module-entering field, performing K-means clustering on the corresponding reconstruction data for preset times respectively to obtain corresponding clustering results, and determining the credit rating of the enterprise according to the clustering results.
2. The enterprise credit rating method based on the convolutional self-encoder as claimed in claim 1, wherein the data selection and data cleaning process is performed on the enterprise data to obtain a modelled field and corresponding modelled enterprise data, and specifically comprises:
Deleting the enterprise data with the missing value larger than a first preset threshold value and the corresponding preset fields according to the preset fields and the corresponding enterprise data to obtain the deleted fields and the deleted enterprise data corresponding to the deleted fields;
and selecting a preset number of fields and corresponding enterprise data from the deleted fields according to a pre-stored selection rule to obtain the cleaned fields and the cleaned enterprise data corresponding to the cleaned fields.
3. The enterprise credit rating method based on the convolutional self-encoder as claimed in claim 2, wherein after a preset number of fields and corresponding enterprise data are selected from the deleted fields according to a pre-stored selection rule, and cleaned fields and cleaned enterprise data corresponding to the cleaned fields are obtained, the method further comprises:
selecting a plurality of fields with the correlation larger than a second preset threshold value from the cleaned fields, selecting the fields with the maximum corresponding enterprise data from the fields with the correlation larger than the second preset threshold value for reservation, and deleting the rest fields and the corresponding enterprise data;
Determining fields corresponding to the enterprise data with the missing value larger than a third preset threshold value in the rest fields, carrying out mean processing on the enterprise data with the missing value larger than the third preset threshold value, and supplementing the enterprise data with the missing value larger than the third preset threshold value to the corresponding fields;
and obtaining the in-mold field and the corresponding in-mold enterprise data.
4. The enterprise credit rating method based on the convolutional self-encoder as claimed in claim 2, wherein in the deleted fields, a preset number of fields and corresponding enterprise data are selected according to a pre-stored selection rule to obtain cleaned fields and cleaned enterprise data corresponding to the cleaned fields, and the method specifically comprises:
selecting a plurality of fields with the correlation larger than a fourth preset threshold value from the deleted fields, selecting the fields with the maximum corresponding enterprise data from the fields with the correlation larger than the fourth preset threshold value for reservation, and deleting the rest fields and the corresponding enterprise data;
screening the deleted fields to obtain a first preset number of times fields, wherein the times fields are fields reflecting the statistics times;
screening the deleted fields to obtain a second preset number of time fields, wherein the time fields are fields reflecting the farthest time or the latest time;
Screening the deleted fields to obtain a third preset number of numerical fields, wherein the numerical fields are fields reflecting numerical operation;
screening to obtain a fourth preset number of simple fields in the deleted fields, wherein the simple fields are fields reflecting that numerical operation is not required;
and obtaining the cleaned field and the cleaned enterprise data corresponding to the cleaned field.
5. The enterprise credit rating method based on the convolutional auto-encoder as claimed in claim 1, wherein before inputting the corresponding modular enterprise data into the network model based on the convolutional auto-encoder according to the modular fields, and obtaining the reconstructed data corresponding to the modular fields, the method further comprises:
respectively inputting the corresponding model-entering enterprise data to a network model to be trained based on a convolution self-encoder for training according to the model-entering fields;
according to the training result, adjusting parameters of the network model to be trained based on the convolution self-encoder;
repeating training until the mean square error between the output result of the network model of the convolutional self-encoder to be trained and the input model-entering enterprise data is smaller than a fifth preset threshold;
And obtaining a network model based on the convolution self-encoder.
6. The enterprise credit rating method based on the convolutional auto-encoder as claimed in claim 5, wherein the parameter adjustment is performed on the network model based on the convolutional auto-encoder to be trained according to the training result, specifically comprising:
and adjusting the learning rate of the network model to be trained based on the convolution self-encoder, the network structure and the numerical value corresponding to the feature map.
7. The enterprise credit rating method based on the convolutional self-encoder as claimed in claim 1, wherein the K-means clustering is performed on the corresponding reconstruction data for a preset number of times according to the modulo field to obtain a corresponding clustering result, and the enterprise credit rating is determined according to the clustering result, specifically comprising:
according to the module-entering field, respectively carrying out K-means clustering on the corresponding reconstruction data for preset times to obtain corresponding clustering results, wherein the clustering results comprise: a cluster label, a center value corresponding to the cluster label;
and determining a cluster label with the maximum central value and a module-entering field corresponding to the cluster label, and taking the cluster label with the maximum central value as an evaluation parameter of the corresponding module-entering field.
8. The corporate credit rating method based on convolutional auto-encoder according to claim 7, wherein after determining the cluster index with the largest center value and the modulo field corresponding to the cluster index and using the cluster index with the largest center value as the evaluation parameter of the corresponding modulo field, the method further comprises:
and obtaining respectively corresponding evaluation parameters according to the mould-entering fields, and calculating the respectively corresponding evaluation parameters through a rating algorithm to obtain the enterprise credit rating.
9. An enterprise credit rating device based on a convolutional auto-encoder, comprising:
at least one processor; and (c) a second step of,
a memory communicatively coupled to the at least one processor; wherein,
the memory stores instructions executable by the at least one processor to cause the at least one processor to perform instructions for:
acquiring an enterprise portrait dimension table, and acquiring data according to preset fields contained in the enterprise portrait dimension table to obtain corresponding multiple pieces of enterprise data;
performing data selection and data cleaning processing on the enterprise data to obtain a model entering field and corresponding model entering enterprise data;
According to the model entry fields, inputting the corresponding model entry enterprise data to a network model based on a convolution self-encoder respectively to obtain reconstructed data corresponding to the model entry fields respectively;
and according to the module-entering field, performing K-means clustering on the corresponding reconstruction data for preset times respectively to obtain corresponding clustering results, and determining the credit rating of the enterprise according to the clustering results.
10. A non-transitory computer storage medium storing computer-executable instructions, the computer-executable instructions configured to:
acquiring an enterprise portrait dimension table, and acquiring data according to preset fields contained in the enterprise portrait dimension table to obtain a plurality of corresponding enterprise data;
performing data selection and data cleaning processing on the enterprise data to obtain a model entering field and corresponding model entering enterprise data;
according to the model entering fields, inputting the corresponding model entering enterprise data to a network model based on a convolution self-encoder respectively to obtain reconstruction data corresponding to the model entering fields respectively;
and according to the module-entering field, performing K-means clustering on the corresponding reconstruction data for preset times respectively to obtain corresponding clustering results, and determining the credit rating of the enterprise according to the clustering results.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210158895.6A CN114519529A (en) | 2022-02-21 | 2022-02-21 | Enterprise credit rating method, device and medium based on convolution self-encoder |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210158895.6A CN114519529A (en) | 2022-02-21 | 2022-02-21 | Enterprise credit rating method, device and medium based on convolution self-encoder |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114519529A true CN114519529A (en) | 2022-05-20 |
Family
ID=81599854
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210158895.6A Pending CN114519529A (en) | 2022-02-21 | 2022-02-21 | Enterprise credit rating method, device and medium based on convolution self-encoder |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114519529A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117151867A (en) * | 2023-09-20 | 2023-12-01 | 江苏数诚信息技术有限公司 | Enterprise exception identification method and system based on big data |
-
2022
- 2022-02-21 CN CN202210158895.6A patent/CN114519529A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117151867A (en) * | 2023-09-20 | 2023-12-01 | 江苏数诚信息技术有限公司 | Enterprise exception identification method and system based on big data |
| CN117151867B (en) * | 2023-09-20 | 2024-04-30 | 江苏数诚信息技术有限公司 | Enterprise exception identification method and system based on big data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Hubbard et al. | Safe bets or hot hands? How status and celebrity influence strategic alliance formations by newly public firms | |
| CN109859052B (en) | Intelligent recommendation method and device for investment strategy, storage medium and server | |
| CN109472646A (en) | A kind of financial product recommended method and device | |
| CN107688645B (en) | Policy data processing method and terminal equipment | |
| US20220343433A1 (en) | System and method that rank businesses in environmental, social and governance (esg) | |
| CN104866484A (en) | Data processing method and device | |
| US20210407010A1 (en) | System and method for providing investment information | |
| JP2022096632A (en) | Computer-implemented method, computer system, and computer program (ranking datasets based on data attributes) | |
| US12072858B2 (en) | Computer-based data collection, management, and forecasting | |
| CN112052385A (en) | Investment and financing project recommendation method and device, electronic equipment and readable storage medium | |
| KR20220034134A (en) | Analysis of intellectual property data about products and services | |
| CN114519529A (en) | Enterprise credit rating method, device and medium based on convolution self-encoder | |
| Petris et al. | Bubble tests in the London housing market: A borough level analysis | |
| JP6978582B2 (en) | Forecasting business support device and forecasting business support method | |
| CN109582476A (en) | Data processing method, apparatus and system | |
| CN108874762A (en) | Online display system and method are reported in a kind of investment research | |
| CN116977091A (en) | Method and device for determining individual investment portfolio, electronic equipment and readable storage medium | |
| US10872376B2 (en) | Systems and computer-implemented processes for occupational risk assessment | |
| Thakkar et al. | Complex Proportion Assessment Method (COPRAS) | |
| Metaxas et al. | A literature review on the financial determinants of hotel default | |
| Olimovna | Analysis of digitalization of the economy of the Republic of Uzbekistan | |
| US20230245235A1 (en) | Cross-functional portfolio database management systems and methods | |
| CN115795289B (en) | Feature recognition method, device, electronic equipment and storage medium | |
| Boda et al. | Prediction of insolvency of Hungarian micro enterprises | |
| Malic et al. | Instrument Development For Measuring Factors Influencing Fog Computing Adoption Based On Quality Of Results (Qor): Content Validity |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |